Specific Aim 1: Investigate protein-coding and non-coding genes as biomarkers and therapeutic targets for cancer. Expression of cancer-related protein coding genes and non-coding miR-21 can be used as robust prognostic classifiers for TNM stage I lung cancer. We analyzed microarray data from 148 stage I lung adenocarcinoma patients to find cancer-related genes associated with prognosis in early stage cancer. Increased BRCA1, HIF1A, and XPO1 and decreased DLC1 were each significantly associated with poor prognosis of stage I-II lung adenocarcinoma in 3 cohorts. For validation, we examined 3 large, publically available, microarray datasets. The 4 coding gene classifier is significantly associated with prognosis in stage I lung adenocarcinoma in each of the cohorts. We are evaluating the combination of the 4 coding gene classifier with non-coding miRNA miR-21 expression as a predictive classifier for stage I lung cancer in a prospective study at UMD. miRNAs are predictive biomarkers of therapeutic outcome for colon cancer. We performed a miRNA profiling study on colon cancer. miRNAs are associated with prognosis and therapeutic outcome of colon adenocarcinoma. Increased miR-21 expression was associated with worse cancer-specific mortality and worse response to 5FU-based chemotherapy. Higher than median miR-21 expression in tumors was associated with poor survival in a Japanese cohort independent of clinical covariates. Therefore, miR-21 is a promising biomarker to identify patients with poor prognosis that may benefit from adjuvant chemotherapy. We will examine global miRNA expression profiles of tumor and noncancerous tissue of 250 colon cancer patients in a prospective predictive study. The integration of diverse genomic data types provides a framework to delineate the molecular characteristics of early stage lung cancer associated with micrometastases. Efforts are focused on the discovery and validation of biomarkers that can aid diagnosis and clinical management of patients with early-stage lung cancer. Tumors which recur have different miRNA and mRNA expression patterns compared to those that do not. We are conducting a multi-platform genomic screen of paired early stage lung cancer and non-tumor adjacent tissues, aimed at defining the molecular differences between aggressive, early stage tumors and those that can be cured with surgical resection. This screen will complement existing mRNA, miRNA and metabolic profiles, and will encompass DNA-seq, RNA-seq, miR-seq and methylation profiling of primary lung tumors from patients with and without disease recurrence. Genome-wide DNA methylation analysis. We are utilizing high-throughput technology to identify changes in methylation patterns associated with early stage NSCLC, clinical and pathological biomarkers, and cancer subtypes. Tumor and adjacent non-tumor tissue DNA were analyzed using Illumina assays, featuring genome-wide coverage of CpG sites. A global increase in methylation was observed in stage I tumors. We found that 7 miRNA loci were hyper-methylated in tumors, including 4 miRNAs located within homeobox gene clusters, miR-196b (HOXA9), miR-615, miR-10a, and miR-10b. We are examining the concordance of miRNA-associated methylation markers in a lung cancer cohort comprised of 76 matched tissue pairs of lung adenocarcinoma from Japanese patients. Preliminary analysis of survival in both cohorts indicates that increased methylation of HOXA9/miR-196b is associated with worse prognosis in stage I, lung adenocarcinoma. MiRNA expression is altered in IBD and may contribute to IBD-associated cancer. We used NanostringTM assays to measure global miRNA expression profiles of 41 surgically resected CD and 50 UC tissues. 35 miRNAs were altered in involved CD and 76 miRNAs altered in involved UC, consistent with a role for miRNAs in IBD. We profiled miRNA expression patterns of 68 paired cancerous and noncancerous tissues from sporadic colon cancer and 24 IBD-associatedcancer tissues. MiR-155 was increased in IBD-associated tumors, but not in sporadic colon tumors. MiR-135b and miR-21 expression were associated with active inflammation in UC and were the top 2 elevated miRNAs in sporadic and IBD-associated cancer. Sporadic tumors had different expression profiles than IBD-associated tumors, suggesting that the different etiology of these tumors results in different miRNA transcriptomic expression patterns. We are investigating a functional role for these miRNA in chronic inflammation and cancer, including the role of miR-135b and miR-155 in both inflammation and IBD-associated cancers. IBD-associated tumors have different and more frequent DNA mutations than sporadic colon tumors. The different etiologies of sporadic vs. IBD-associated colon cancer suggest that the accumulation of somatic mutations in each of these diseases will be quite different. In collaboration with Bert Vogelstein, we will use next generation DNA sequencing of colon tumors that have arisen in the context of UC or CD. This study is designed to compare the types and frequency of somatic mutations of UC- and IBD-associated tumors with sporadic tumors and will represent the first attempt to compare the somatic mutations and chromosomal rearrangements in this way. Specific Aim 2: Metabolomic profiles are diagnostic and prognostic biomarkers for lung cancer Urine metabolites are diagnostic and prognostic biomarkers of lung cancer. Mass spectrometry-based untargeted metabolic profiling of urine samples obtained from 469 lung cancer patients and 536 healthy population controls has been done. We identified 4 metabolites which are associated with both lung cancer diagnosis and poorer survival. A combination of the 4 biomarkers resulted in stronger associations with prognosis, suggesting that they may be independent of one another. Validation experiments confirmed that these metabolites were significantly higher in an independent sample set of lung cancer cases. The metabolome of 62 tumor and 62 adjacent nontumor tissues was profiled, linking two urinary biomarkers directly to tumor cell metabolism. Ongoing steps for this study include testing urine specimens from the National Lung Screening Trial. We are testing whether these biomarkers are predictive of lung cancer risk in a nested case-control study within the prospective Southern Community Cohort Study. We will profile blood serum and plasma from the same donors and combine these results with urine metabolites with the goal of improving diagnostic and prognostic tools. Specific Aim 3: Identify bacterial communities associated colon cancer The mucosa-associated microbiome identifies colon cancer patients We designed experiments to test 3 platforms for microbiome analysis, 454 pyrosequencing; MiSeq and PhyloChip. Using PhyloChip analysis, we did a study of colon tumor and adjacent noncancerous tissue. Bacterial communities and specific bacteria associated with CRC were identified and distinguished CRC from controls. When we compared the bacterial abundances between specimens from healthy controls vs. CRC, 17 bacteria species were found to have significant differences. The majority of the differences came from bacteria in the class gammaproteobacteria and alphaproteobacteria. When we sub-divided these abundances into representative taxa by phylotype, we saw that the difference between tumor and nontumor was small and the majority of samples were dominated by Proteobacteria, Planctomycetes, and Bacteroidetes. There was loss of 4 species of Pseudomonas in AAs compared to Caucasians. We will perform 16S rRNA sequencing analysis on these pilot samples, and determine the best platform on which to sequence a larger set of CRC to study microbial communities associated with CRC and investigate potential differences among ethnic groups.